Re: NeuronDB RDF and OWL

> We just finished exporting the NeuronDB of Senselab into RDF and OWL.
> 
> http://neuroweb.med.yale.edu/senselab/


Great!

Here are some thoughts I had while browsing through the OWL version (I post them in public so others do not give you redundant feedback):

According to the Pellet reasoner, the ontology is fully consistent.

I guess you plan to change this in the final release anyways, but all the classes need to have a value for the rdfs:label property. The URIs are not intended to be read by humans, so none of the classes has an actual name in the current version of the ontology.

Assuming that the class names will be identical with the current URIs:

“Forebrain” should be renamed to “Prosencephalon”, to be consistent with the naming of the other major brain regions (e.g. “mesencephalon”, “metencephalon”). Of course, you could also rename the other classes to the English trivial names (“midbrain” etc.). 

With the current names of the classes, many of the class-subclass relations are in fact whole-part relations (e.g. the ontology states that “Vestibular_Organ” is a subclass of “Cochlea”, but in fact it is a part of the cochlea). If you want to preserve the current class structure, but be ontologically consistent, I would suggest to rename such classes from “X” to “X or part of X”. For example, “Vestibular_Organ” would become “Vestibular_Organ_or_part_of_Vestibular_Organ”. Now it would be correct to state that “Cochlea” is a subclass of “Vestibular_Organ_or_part_of_Vestibular_Organ”, because the superclass includes parts.

The IDs/names of the subclasses of “Dendritic Compartment” and “Axon” could be a little clearer (I assume that d, m and p stand for distal, medial and proximal, AH is axon hillock and T is terminal?).

Some of the subclasses of “Neuron_Receptor” and “Neuron_Transmitter” (Neurotransmitter) have names in the plural form (e.g. “Ion_Receptors”). This should be changed to singular to be consistent with the rest of the ontology.

Using Protege 3.2, I have troubles view the annotations that relate facts to individuals belonging to the "Notes" class. Looking at the source, I see that rdfs:seeAlso and blank nodes were used. I guess it is Protege's fault that these relations are not displayed.


cheers,
Matthias Samwald

-- 
"Feel free" - 10 GB Mailbox, 100 FreeSMS/Monat ...
Jetzt GMX TopMail testen: www.gmx.net/de/go/mailfooter/topmail-out

Received on Tuesday, 13 March 2007 12:48:18 UTC